A Kjeldahl crude protein analysis takes four to six hours. An NIR scan of the same ingredient takes less than one minute. This time difference is not just operational: it changes what is possible to do with analytical data within a feed mill's production cycle. Lots that previously waited for results before release can be analyzed during receiving. Formulations that depended on composition tables can now be adjusted with real data from the incoming raw material.
Near infrared spectroscopy is not new to the food industry. But its use in animal nutrition still carries many questions: what NIR actually analyses well, where it fails, which type of equipment makes sense for each operation, and how to integrate results into the formulation process. This article answers these questions in a technical and direct manner.
How NIR works: the physical principle and the role of calibrations
NIR operates in the spectral range between 800 nm and 2500 nm of the electromagnetic spectrum. When an infrared beam strikes a sample, part of it is absorbed and part is reflected. The molecular groups present in the sample — C-H, O-H, and N-H groups, which form the basis of proteins, lipids, carbohydrates, and water — absorb energy at specific wavelengths. The resulting spectrum is essentially a chemical fingerprint of the sample.
The equipment captures this spectrum in seconds. But the raw spectrum has no analytical value on its own. For a crude protein or metabolizable energy value to be extracted from an NIR scan, a calibration model is needed: a set of mathematical equations developed from samples analyzed simultaneously by NIR and by reference chemical methods (Kjeldahl for nitrogen, Soxhlet for ether extract, Van Soest method fiber analysis, among others).
Calibration: what defines the quality of an NIR result
Calibration is the most critical component of any NIR system. High-cost equipment with poor calibration produces imprecise results. Simpler equipment with good calibration can be more reliable. Calibrations can be global — developed by equipment suppliers with samples from various regions worldwide — or local, built with samples from the operation itself.
Global calibrations work well as a starting point, but rarely capture the full variability of ingredients from a specific region or particular supplier. For ingredients with high regional variability, such as soybean meal from different crops or animal-origin meals, local calibrations or adjustments to global calibrations tend to deliver more precise results. Another frequently underestimated aspect is continuous model maintenance: as new suppliers are added, crops change, or ingredient production processes are altered, calibrations need to be revised and updated.
What NIR analyzes and with what reliability
The usefulness of NIR depends directly on which parameter is being analyzed. It is not correct to say that NIR "analyzes everything," nor that it completely replaces wet chemistry. The practical distinction is as follows: NIR performs excellently for parameters that vary with the molecular composition of the sample and that have good spectral representation in the near-infrared range. For other parameters, results are less reliable or unfeasible.
Parameters with good NIR performance in feed ingredients
For corn and sorghum, the parameters with best NIR performance are moisture, crude protein, and ether extract. Prediction of metabolizable energy from these components is also well established. For soybean meal, crude protein and moisture have widely available calibrations with good precision; soybean processing variability between plants may require calibration adjustment by origin.
Animal-origin meals — meat and bone meal, feather meal, fish meal — show greater compositional variability and require more robust calibrations. When well calibrated, NIR can predict crude protein, moisture, and ether extract with precision acceptable for formulation use. Prediction of metabolizable energy for animal meals is technically possible, but requires special attention to the calibration sample set, which must adequately represent the ingredient's actual variability.
Where NIR has limitations that need to be known
Mycotoxins are not reliably detected by conventional NIR. Aflatoxin, fumonisin, and zearalenone are present at concentrations in the parts-per-billion range — far below the analytical sensitivity of standard NIR spectroscopy. NIR systems specific to mycotoxins exist but operate on different principles with higher costs. For mycotoxin control, conventional chemical analysis or immunoenzymatic kits (ELISA) remain the appropriate methods.
Individual minerals also have limitations. Calcium and total phosphorus can be estimated with specific calibrations for certain ingredients, but precision tends to be lower than for organic parameters. Individual amino acids, such as digestible lysine and methionine, can be predicted by NIR in soybean meal and corn with adequate calibrations, but prediction errors tend to be larger than for total crude protein, and using these values for fine formulation adjustment requires careful validation.
Types of NIR equipment and where each applies
The choice of NIR equipment type depends on where in the operational chain the analysis needs to happen and what level of precision is required for the decision to be made with the result.
Benchtop NIR
Benchtop equipment is installed in laboratories and offers the highest analytical precision within NIR technologies. It is suitable for analyzing samples arriving at the plant's central laboratory — raw material receiving, process control, and finished product verification. The typical routine involves minimal sample preparation (homogenization, grinding in some cases), scanning, and result recording. Integration with laboratory management systems enables automated flow of analytical data to other operational systems.
Portable NIR
Portable equipment enables analysis at the sampling point, without needing to transport the sample to the laboratory. They are useful in operations with multiple plants, direct analysis in silos or receiving trucks, and situations where time between sampling and result must be minimized. Precision tends to be slightly lower than benchtop equipment, and calibration management across multiple devices requires a specific protocol.
Inline NIR
NIR sensors installed directly on conveyors, screws, or pipes allow continuous monitoring during the production process, without manual sample collection. They are applied mainly for real-time moisture monitoring, detection of composition variations during mixing, and automatic alerts when material deviates from expected parameters. Implementation is more complex and requires greater infrastructure investment, but delivers the highest level of process control possible with NIR technology.
From spectrum to formulation software: the data flow that closes the cycle
The strategic value of NIR in a feed mill is realized when the analytical result ceases to be a number printed on a report and begins to directly feed formulation. This flow requires integration between three systems: the NIR equipment, the LIMS (laboratory management system), and the formulation software.
The LIMS receives NIR results — automatically or via structured import — and links them to the corresponding raw-material lot. With analysis history organized by ingredient and supplier, the system can calculate moving average compositions, identify variation trends over time, and alert when a specific lot shows values outside historical norms. These updated composition data are then made available to formulation software, which can use the actual values of the lot in stock rather than table values, reducing the gap between formulated and produced feed.
When this cycle is working, a raw material with crude protein below expectations does not need to be automatically discarded or used without adjustment. The formulator receives the real data, evaluates the impact on formulation cost and nutrition, and decides — with information, not assumption.
Labinfy is a cloud-native laboratory platform designed for industrial animal-nutrition laboratories, with integration to analytical equipment, batch traceability, audit trail, and statistical process control in a single environment.
Implementation: what plants need to evaluate before investing
The decision to implement NIR involves variables beyond equipment selection. Before any investment, it is worth structuring the evaluation around four practical questions.
First, what is the current analysis volume and frequency? Plants performing few analyses per month have less operational pressure to justify investment than operations with high raw material receiving volume. Second, what calibrations are available for the ingredients used? It is essential to verify whether the equipment supplier or external partners have adequate calibrations for the operation's ingredient mix, especially less common alternative or regional ingredients.
Third, what will the NIR plus wet chemistry policy be? NIR does not eliminate the need for reference analyses. A well-defined policy establishes when to use NIR as a screening method, when wet chemistry confirmation analysis is mandatory, and how to handle discrepancies between the two methods. Fourth, who will be responsible for calibration maintenance? This is the most underestimated aspect of NIR implementation. Without active maintenance, calibration models become outdated and result precision progressively deteriorates without the team necessarily noticing.
The real cost-benefit of NIR in a feed mill operation
Return on NIR investment has three main components. The first is the reduction of formulation gap: when the plant formulates with composition values that do not reflect the actual ingredient being processed, the final product may have excess or deficiency of nutrients. Excess represents wasted cost; deficiency represents animal performance risk. NIR integrated with the formulation system reduces this gap.
The second component is the reduction of non-conforming lots. Out-of-specification ingredients detected at receiving prevent the problem from advancing to the production process. The cost of reprocessing or discarding a finished feed lot is significantly higher than the cost of rejecting an ingredient at receiving.
The third component, frequently ignored in calculations, is the increase in analytical capacity without proportional team increase. With NIR, it is possible to analyze all lots of a critical raw material instead of working with statistical sampling due to laboratory time constraints. This improves quality control coverage without necessarily hiring more analysts.
The sum of these three factors usually justifies investment in medium and large-scale operations. In smaller operations, analysis must consider the ingredient profile used, the criticality of quality deviations on animal performance, and the possibility of sharing equipment between plants or outsourcing NIR analysis as a service.